.png){kind=icon}
.png){kind=icon}
Return
GenMix: Effective data augmentation with generative diffusion model image editing
.png){kind=icon}
OA
AI
DOI：10.1016/j.eswa.2026.132273.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Data augmentation is widely used to enhance generalization in visual classification tasks. However, traditional methods struggle when source and target domains differ, as in domain adaptation, due to their inability to address domain gaps. We introduce a novel GenMix, a generalizable prompt-guided generative data augmentation approach that enhances both in-domain and cross-domain image classification. Our technique leverages image editing to generate augmented images based on custom conditional prompts, designed specifically for each problem type. By blending portions of the input image with its edited generative counterpart and incorporating fractal patterns, our approach mitigates unrealistic images and label ambiguity, improving performance and adversarial robustness of the resulting models. Our GenMix further incorporates a faithfulness-aware filtration step that retains only semantically consistent edited images for augmentation. Efficacy of our method is established with extensive experiments on eight public datasets for general and fine-grained classification, in both in-domain and cross-domain settings. Additionally, we demonstrate performance improvements for self-supervised learning, learning with data scarcity, and adversarial robustness. As compared to the existing state-of-the-art methods, our technique achieves stronger performance across the board. The Code is available on https://github.com/khawar-islam/GenMix
Keywords:
Data augmentation
Diffusion models
Generative models
Synthetic data
Image editing
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
7.5
Papers:
2.9W
Citations:
10.2W
View more organizations